A main focus of systems neuroscience is to understand how sensory information is encoded and used to guide behavior. Perceptual decision-making, like nearly all normal behavioral processes and disorders of the nervous system, is thought to involve the activity of large groups of neurons. Technical limitations, however, have forced most physiological studies to focus on single neurons. These studies have provided many important insights, but they necessarily miss key information about the relationship between groups of sensory neurons and decisions. For example, single neuron responses cannot tell us how that neuron's activity interacts and is combined with that of other neurons within or between cortical areas. Furthermore, my prior work showed that the cognitive state of even a well-trained subject fluctuates greatly from moment to moment, with striking consequences on performance on perceptual tasks. Therefore, combining information from single neurons recorded at different times produces an average over many behavioral states. The experiments in this proposal use the activity of many simultaneously recorded neurons while animals perform a visual discrimination task to track each perceptual decision while it is in progress. Our goal is to understand how the responses of neurons with different physiological, visual and cognitive properties who come from different cortical areas and interact with nearby neurons in different ways are combined to drive decisions. In Specific Aim 1, we will ask how the responses of neurons with different functional properties are combined to guide decisions by assessing the relationship between the animal's choices and the activity of neurons that differ in their ability to encode the relevant information or are modulated differently by cognitive factors such as visual attention. In Aim 2, we will assess the validity of current models of decision-making by determining the effect of shared (or correlated) response variability on perceptual performance. The responses from a few dozen simultaneously recorded neurons will provide a means of assessing the correlation structure of a large population and isolating the effects of correlated variability from accompanying changes in firing rates. In Aim 3, we will examine the role of neurons in different visual areas in decisions, ask whether there is complementary choice-related activity in different areas, and determine whether the role of each area is flexible We will record simultaneously from groups of neurons in V4, a ventral stream area, and MT, a dorsal stream area, to determine whether we can better predict the animal's decisions from groups of neurons in the two areas together or from one area alone. We will also adjust the visual stimulus to favor the tuning properties of neurons in each of the two areas to see whether the role of each area in decisions depends on the suitability of its neurons for the particular perceptual task. We hope that extending the study of decision-making to populations of neurons will improve our understanding of the underlying neuronal mechanisms and guide future experimental and theoretical work.